gen‑ai.news

Demo archive

Every past weekly demo pick, newest first.

  • Krea 2 Turbo bundles text-to-image generation and image editing into a single Gradio interface, drawing on multiple model variants including a dedicated identity-edit model and several LoRA options.

  • A collection of Qwen Image Edit LoRAs is demoed here through a Gradio interface, letting you run targeted image editing tasks across styles including anime conversion, manga tone, noir comic panels, Pixar-inspired 3D, polaroid photo, upscaling, and multi-angle generation.

  • A demo interface for testing a curated collection of Qwen Image Edit LoRAs, pulling together over two dozen specialized models covering relighting, face swapping, pose control, skin detailing, upscaling, and style transfer into a single Gradio-based tool.

  • Omni Video Custom-fast-motion is a Gradio-based tool that accepts text prompts or images as input and generates video output, with support for extending existing video clips.

  • MiniMax H310 Aug 2026

    MiniMax H3 generates video alongside a synchronized soundtrack, combining MiniMax's H3 model with an audio component to produce video and sound together in a single output.

  • Mage-VL3 Aug 2026

    Mage-VL runs Microsoft's 4B parameter multimodal model directly in a Gradio interface, handling both image and video understanding through codec-native processing.

  • A Gradio interface wrapping LTX 2.

  • Wan2.

  • Pro Realism Edit Studio is an image editing tool built by Sneak-Moose that accepts one or two input images and applies edits using a stack of several Qwen-based vision-language models.

  • A collection of Qwen Image Edit LoRAs is assembled here into a single Gradio interface, letting you apply named styles like Noir Comic Book Panel, Pixar-Inspired 3D, Polaroid Photo, Midnight Noir Eyes Spotlight, and Manga Tone to images through a common editing pipeline.

  • Omni Image Editor bundles several common image manipulation tasks into a single Gradio interface: text-to-image generation, image editing, upscaling, and watermark removal.

  • Ideogram 48 Jun 2026

    Ideogram 4 runs state-of-the-art open-weights image generation through a Gradio interface, offering both fp8 and nf4 quantized model variants alongside a vision-language component built on Qwen3-VL.

  • Bonsai Image WebGPU runs image generation entirely in the browser using WebGPU, drawing on 1-bit and 2-bit quantized 4B-parameter models to keep compute requirements low enough for client-side inference.

  • Wan2.

  • Pixal3D18 May 2026

    Pixal3D converts a single image into a 3D asset by combining depth estimation, background removal, and mesh generation through a pipeline that draws on models from Microsoft, Meta, and Black Forest Labs.